Whitelight Diode

[steve-o]

Here's the best I can do today (tired, wishing re-tired)
No soft start yet. Do we really need soft start? The LM3XX series regulators are pretty stable I believe. The analogue modulation maybe could be done with VCO in /PWM out from the oscillator to control pulse width hence brightness. Like pin 5 on a 555 timer. (I wish we could use a 555-too slow of course.) I know your idea had the fet across the LD. Can you put up a schematic of that?

[The_Doctor]

I'm tired too... Check my posts when awake again, they cover everything I can think of so far. I can't add any schematic details that aren't already in that diagram here.

Your series modulator won't be safe, when it's not conducting, little current will flow, and in an effort to make 1.25V across its sense resistor, the regulator will raise the voltage. That could pop the LD during the fast rise to conduction during the few microseconds the regulator takes to respond. Shunting is the way to be safe. Even then, it needs care to avoid stray inductances and capacitances when switching at those high speeds, to avoid LD-destroying peaks, hence the clamp diodes as main method of protection in that schematic.

I like the PWM method for analog mod. Normally I'd want to avoid it to save diode stress, but in this case it could work, BUT, the full output must still be 50% or whatever the chosen DVD diode wants for max pulse drive. Any square wave osc with a filter to make triangle will work, feed it to an open loop op-amp with positive feedback like I said, it's cheap and easy, and makes sure that hysteresis prevents unwanted switching, and also accelerates the fast changes for neat pulse shapes.

My driver idea will be an opto-FET driving the IRF630 shunting an LD fed by a fixed current. It's brutally simple, but with the memcap soft start on supply input offers most of the advantages that seriously expensive drivers offer. With the addition of a few zeners and other diodes, it should be almost impossible to overdrive the mod input or the supply input with anything but brute carelessness or deliberate sabotage. I haven't made a neat schematic yet, I'm still waiting for opto-FET's. I won't go beyond back-of-envelope stuff till I have a working circuit built. It won't have the fast pulse drive but that could be added to it, all the parts used should work fast enough.

[steve-o]

OK point taken..
He's using the FET in linear mode? Alot more heat generated that way.
He said that the circuit is optimized for a 5mW diode.
Also I thought that we were trying to get away from using the regulator in a voltage mode and go with current mode.
If the 7805 failed, which is RARE from what I've seen the zener would pop with what we're using --actually we cant use a 7805- it's limited to 1 amp.
Welp time to pack it in and go have a toddy
How 'bout this: (I dont know if the 338 is gonna like being shorted at 2 MHz though)

[The_Doctor]

OK point taken..
He's using the FET in linear mode? Alot more heat generated that way.

Constant current, remember? The only extra heat the system puts out during shunt is that which is not put out in a laser beam during not-shunt. Maybe 50% extra, assuming LD conversion efficiency is about 33%. It's why I chose a TO220 packaged MOSFET, it will definitely cope, clamped to the chassis.

He said that the circuit is optimized for a 5mW diode.

Even those take 40 to 70 mA drive, this isn't so different, within a factor of five, is just resistor tweaks, mostly, and semis that can be clamped to a chassis for heatsinking.

Also I thought that we were trying to get away from using the regulator in a voltage mode and go with current mode.

Yes, we are. But if you had a series FET raising a high resistance, the current will fall, and to try to raise it, the regulator MUST try to put that 1.25V across its sense resistance, cos that's how it knows to regulate anything. This means that the instant the FET resistance drops, the volts trying to overcome its previous resistance will now slam current through the LD until the regulator can catch up. It will, in a couple of microseconds, but this could still be slow enough for COD to happen.

If the 7805 failed, which is RARE from what I've seen the zener would pop with what we're using --actually we cant use a 7805- it's limited to 1 amp.

I was thinking that, too. But zener diodes can be butch wee beasties, it could be a 5 amp zener or more, and anyway, he's current limited his main supply with a resistor for filtering it. I like that idea, never tried that before, should have. I've done it with 1R after the current regulator (also good for test point to read current on a voltmeter, as he does), but using an even bigger resistance in the main input is a good way to allow basic wall warts especially cheap SMPU's as power source. That resistor and the following large cap should stop any serious nonsense from reaching the regulator let alone the LD.

I'm curious that he decided his 7805 was accurate enough. That said, my recent test of an LM2904 LDO regulator was less than eddifying, its load regulation is not cool. An LM317 would be better, I might use one, it's just a couple of fixed resistors, and no caps, it's stable as is. I'd lose my 5V input capability, but that's not this business, it was for the power meter I mentioned elsewhere.

I see what you mean about current not voltage now... His schematic.. I think he chose that so soft start would be easier to design. I'll go with constant current though, and manage the soft start by controlling the MOSFET gate voltage, I think. Either that, or the memcaps. But I hate paying for memcaps.

Welp time to pack it in and go have a toddy
How 'bout this: (I dont know if the 338 is gonna like being shorted at 2 MHz though)

Better, but consider the input. It ought to have differential input and +5V for full output. If it has that it's matching ILDA spec and can take analog or digital. I haven't yet chosen between fast pulse and PWM for brightness control, or actual analog. I think analog though, I chose parts for high isolation, high linearity, so it should be safe and easy. An opto-FET makes a nice way to get isolated differential inputs with extremely high noise immunity. That it will be fast enough for multi-MHz TTL also will be an extra. It would demand lower capacitance protection across the LD that pure analog, but the isolation and input protection might be good enough to support the risk. Bottom line is: average power is average power. If I take a Rohm diode and give it 50% at 240 mW, that's no more than 120 mW CW, and I can get a tad more than that in CW anyway, so I'll go with pure analog, it's simpler. It will be good for up to 50 KHz or more though, I think that is a good value to limit to.

Re shorting the regulator, it should be fine. Just got to make sure that your input voltage doesn't overheat it when it's driving a short. This is one reason why ganging LM317's might beat an LM338. That's still cheap, and it's easier to get rid of heat from them. I like to allow a wide range of input voltage, ideally between 5V and 15V as supply, so it's esy to find a source, and harder to destroy the laser.

[steve-o]

Food for thought-

From Sam's FAQ:

The difference between threshold and maximum current is usually quite small; no more than 10% or 20% of the threshold. The threshold current varies greatly from one device to another (even within the same type number) and also varies with temperature. Result: setting a fixed current value is doomed to failure. For some lasers, and on some days, it will be under the threshold and no laser action will occur; on other days, it will be over the maximum current and your precious laser will turn into a useless LED (like the original posting in this thread). The only safe way is to use the monitor diode current to servo the light output. Even this isn't ideal because the monitor current is different for different lasers, but:

• It doesn't vary significantly with temperature;
• Many laser manufacturers give you a test sheet with each device stating the actual monitor current for full output;
• You can provide an adjustment anyway.

But BE CAREFUL. Transient overdriving, even for very short times, can seriously damage the lasers. Transients commonly occur:

• because your feedback circuit rings (or worse. oscillates) so that the drive current occasionally exceeds the maximum.
• because of PSU on/off transients.
• because you have used a socket for the laser, and the photodiode connection is flaky: if it comes disconnected, your feedback circuit will think there isn't enough drive to the laser and will crank up the current to destruction level.
• because you are trying to modulate the laser brightness with some AC signal and either you overdo it, or the feedback circuit overshoots.
• because you have a pot. somewhere in the circuit to adjust for full output, and its wiper is noisy.

Above all, remember that it is excessive light output that destroys lasers. The heating effect of the drive current is not a big problem except that it has the effect of pushing the threshold current down. Excessive light levels, on the other hand, can damage the tiny end mirrors of the lasing crystal.

Sharp (one of the big suppliers of laser diodes) also make some nifty 8-pin drive chips that are pretty good if you don't need to modulate the laser rapidly. For modulation, consider setting the light output close to 50% of full output using a really slooooowww-responding feedback circuit, and then impressing a fixed-amplitude modulating current on the laser. This is OK because the gradient of the light/current graph is reasonably predictable for any given laser type, so it's possible to calculate a suitable safe modulating current from the data sheet. Good luck to all - and don't forget the eye safety regulations.

This is neat(also from Sam's)

In +-------+ Out R1 .25 ohms
Vin o----| LM338 |-------/\/\/----+-----o 0 to 5 A current source
+-------+ |
| Adj. +----+
| cw | |
| 1K ^ / _|_,
+-------------->\ '/_\ LM385-1.2
/ |
| |
+----+
|
+------------------------+
| I = 0.5 to 1.5 mA sink |
+------------------------+
_|_
-

OOPs --^that didnt come out too well
See attachment on bottom

...Let me offer the following simple circuit, which I just created and haven't tried but 'oughta work' as a solution to this problem.
By contrast, this circuit can only *decrease* the current from the 1.25V/R value, but it easily handles a 10:1 range (or even much more) and the voltage across the sense resistor is never more than 1.25V, allowing low supply voltage (e.g. 5 V) and keeping the dissipation low.
The 1K pot selects a portion of the floating 1.23 V reference voltage, and tricks the LM317 or LM338 into correspondingly reducing the voltage across the 0.25 ohm current-sense resistor. The pot is conventional and may be panel mounted. It should be possible to nearly shut off the LM338 (a minimum quiescent current will still flow). The current sink, I, which powers the floating 1.23 V reference, is not critical and may be a simple current mirror (sorry to see the TL011 gone!), or even a resistor to ground or any available negative voltage, depending upon the desired current-source voltage-compliance range. That's it!

[The_Doctor]

Beware. While Sam's LaserFAQ is full of good stuff, it's also full of OLD good stuff! Things change. If a laser diode cannot be run in constant current more, and not even easily in constant power mode, how exactly do we explain the absence of photodiodes in most if not all high power single mode diodes made recently?

That circuit is Winfield Hill's, it's the one I mentioned earlier, that uses a bandgap and a constant current source to work right. I haven't got those parts so I never built it.

The heating effect of the drive current is not a big problem except that it has the effect of pushing the threshold current down.

If true, the threshold current can NOT be used as a guide to maximum safe current, because the maximum safe current rises with temperature, not falls. This needs examining, because many here have said that threshold IS a good guide to safe maximum.

Anyway, I have no problem with constant current drive, I never used anything else, and once I found the safe limit for a diode type, I never saw one die from overcurrent. Once you find a value that is safe, long-term, it is usually safe over temperatures between 0C and 40C or more, higher limit set by maker's specs. I usually do my testing at around 15C, not more than 20C, and only test for freezing temperatures when I've managed to find safe life of >2000 hours at room temp.


EDIT:
For ASCII art, try CODE or PRE tags in UBB code...

[yaddatrance]

Yes, you won't find a PD on any of the big diodes, You have to run in
constant current mode. Sam's is right in that a passively cooled diode
can run away and do bad things, but that's why you have TEC drivers
to maintain the environment so the diode is stable. The nicer ones have
bipolar drivers so it can HEAT as well as cool.

A "soft start" is nothing more than an R-C circuit on the main DC power.
Determine the current it needs, set the resistor accordingly and find a
capacitor to match.

The key thing for fast modulation is to keep the current when off at
just below the threshhold for lasing. Note that you will not see
femtoseconds using traditional discrete components as your chip will have
to operate in the hundreds of terahertz range! Femtosecond lasers
are made using creepy optics math.

Now the ONLY reason that femtosecond lasers are used is that is
the most convenient way to generate the power required to create
a super-continuum. If you have "enough" power, you can do it in
the atmosphere without even a fiber...

http://www.dodsbir.net/Sitis/archive...Bookmark=28981

Now, I want THAT for my laser show

[steve-o]

Yadda- Cool! I guess audience scanning is out with this setup?

Dr- Dv/dt was my question about the FET. I'm not sure if we're talking linear driving of the gate or PWM? The latter is easier on the power device (less heat.)

Steve

[The_Doctor]

A "soft start" is nothing more than an R-C circuit on the main DC power.
Determine the current it needs, set the resistor accordingly and find a
capacitor to match.

Agreed, I think that this seems to beat the more elaborate suggestions. I've about decided that for testing the bluray diode this is the only way I can fit into the assembly I'll be using, too.

Re femtoseconds, true, I'd allowed myself to drift way off topic.. I guess femtosecond lasers will be way out of my league for as long as I can imagine. The only thing I can think of about them is two things: 1. I guess they use some kind of passive q-switch to work that fast. And 2. The idea of a supercontinuum in a fibre reminds me of someone hitting a telegraph wire with a stick. The harder and faster the stick, the brighter and more harmonically rich the sound returned. Maybe a lousy analogy but I like it.



Steve-o, I'd go with analog mod pure and direct, if linearity allows. The current is diverted but still constant, it's never greater than the diode sees, though the power dissipated as heat is maybe 50% more cos none is being cast as light by the FET. While PWM is gentle on the FET, analog is gentle on the diode, which is where gentle treatment is really needed.

[steve-o]

So we're not talking about pulsing the LD anymore- as was the original subject of this discussion?